MysteryX
16th July 2016, 04:32
I'm still working on my script to upscale 288p VCD into 768p videos. Something I just changed is to tune the KNLMeans denoiser way down as it was blurring the picture too much and the SuperRes algorithm update makes it softer and more tolerant to noise. Plus if I'm applying 2 denoisers, they have to be applied more softly.
I apply the work in this order: KNLMeans, Double, SMDegrain, InterFrame, Double.
Inserting SMDegrain at the beginning kills too much detail, and applying it after a frame double does seem to help... but I'm really not sure what it's actually doing. Shapening must be enabled for it to give good results... and what does it do besides that sharpening? Is SMDegrain's operation affected by doing it right after a frame double that "softens" the picture? Sometimes I see details becoming a little bit clearer, sometimes I see it randomly transforming noise, and sometimes I really don't see much difference at all.
So yeah, what does SMDegrain do exactly and how can I make the best use of it in this context?
file="VCD.mpg"
LWLibavVideoSource(file, cache=False)
AudioDub(LWLibavAudioSource(file, cache=False))
Crop(4, 32, -4, -28)
Dither_convert_8_to_16()
Dither_resize16nr(Width, Height/2, kernel="Spline36", csp="YV24", cplaces="MPEG1")
KNLMeansCL(D=2, A=2, h=0.6, cmode=true, device_type="GPU", device_id=0, lsb_inout=true)
SuperResXBR(5, 1, 0, XbrStr=2.7, XbrSharp=1.3, MatrixIn="Rec601", lsb_in=true, lsb_out=true)
Dither_resize16nr(Width, Height/2, kernel="Spline36", csp="YV12")
SMDegrain(thsad=100, prefilter=1, contrasharp=true, lsb_in=true, lsb_out=true)
DitherPost(mode=6)
InterFrame(Cores=8, Tuning="Smooth", NewNum=60, NewDen=1, GPU=true)
Dither_convert_8_to_16()
Dither_resize16nr(Width, Height/2, kernel="Spline36", csp="YV24")
SuperResXBR(3, 1, 0, XbrStr=2.7, XbrSharp=1.3, fWidth=1288, fHeight=784, fKernel="Bicubic", fB=0, fC=.75, FormatOut="YV12", lsb_in=true)
ResizeX(1280, 768, 4, 8, -4, -8)
Prefetch(8)
I apply the work in this order: KNLMeans, Double, SMDegrain, InterFrame, Double.
Inserting SMDegrain at the beginning kills too much detail, and applying it after a frame double does seem to help... but I'm really not sure what it's actually doing. Shapening must be enabled for it to give good results... and what does it do besides that sharpening? Is SMDegrain's operation affected by doing it right after a frame double that "softens" the picture? Sometimes I see details becoming a little bit clearer, sometimes I see it randomly transforming noise, and sometimes I really don't see much difference at all.
So yeah, what does SMDegrain do exactly and how can I make the best use of it in this context?
file="VCD.mpg"
LWLibavVideoSource(file, cache=False)
AudioDub(LWLibavAudioSource(file, cache=False))
Crop(4, 32, -4, -28)
Dither_convert_8_to_16()
Dither_resize16nr(Width, Height/2, kernel="Spline36", csp="YV24", cplaces="MPEG1")
KNLMeansCL(D=2, A=2, h=0.6, cmode=true, device_type="GPU", device_id=0, lsb_inout=true)
SuperResXBR(5, 1, 0, XbrStr=2.7, XbrSharp=1.3, MatrixIn="Rec601", lsb_in=true, lsb_out=true)
Dither_resize16nr(Width, Height/2, kernel="Spline36", csp="YV12")
SMDegrain(thsad=100, prefilter=1, contrasharp=true, lsb_in=true, lsb_out=true)
DitherPost(mode=6)
InterFrame(Cores=8, Tuning="Smooth", NewNum=60, NewDen=1, GPU=true)
Dither_convert_8_to_16()
Dither_resize16nr(Width, Height/2, kernel="Spline36", csp="YV24")
SuperResXBR(3, 1, 0, XbrStr=2.7, XbrSharp=1.3, fWidth=1288, fHeight=784, fKernel="Bicubic", fB=0, fC=.75, FormatOut="YV12", lsb_in=true)
ResizeX(1280, 768, 4, 8, -4, -8)
Prefetch(8)